Data Center Server Racks with Liquid Cooling

Liquid-cooled racks are essential for high-density, AI, and HPC workloads, offering superior heat removal, energy efficiency, and scalability compared to traditional air-cooled systems.

Overview of Liquid-

Cooled Racks

Liquid cooling in data centers uses fluids—typically water or dielectric liquids—to absorb and transfer heat from servers more efficiently than air. This is critical for high-density racks, where power consumption and heat generation exceed the limits of conventional air cooling, often surpassing 15–20 kW per rack and reaching 100 kW or more. Liquid-cooled racks are increasingly adopted for AI, machine learning, and HPC workloads due to their ability to maintain optimal operating temperatures while reducing energy costs.

Types of Liquid CoolingLiquid-to-the-Rack (Rear Door Heat Exchanger)Chilled water coils are installed at the rear of the rack. Hot air from servers passes through the coils, transferring heat to the liquid loop. Suitable for medium-density racks and hybrid cooling setups.

Direct-to-

Chip (Direct Liquid Cooling)Cold plates are attached directly to CPUs, GPUs, DIMMs, and other heat-generating components. Coolant flows through sealed tubes, removing heat efficiently without contacting electronics. Supports extremely high-density racks, often exceeding 100 kW per rack.

Immersion Cooling

Servers are fully or partially submerged in dielectric fluid. Single-phase systems circulate liquid to a heat exchanger; two-phase systems allow the fluid to boil and condense, enhancing heat transfer. Provides maximum thermal coverage and near-silent operation, ideal for ultra-high-density deployments.

Key Components

Coolant Distribution Units (CDUs): Supply and return coolant to racks, often with redundancy and intelligent pump control. Capacities range from 70 kW to over 1 MW per unit. Cold Plates: Interface directly with chips to extract heat efficiently.

Manifolds and Hose Kits: Ensure flexible and scalable coolant distribution within and between racks. Integrated Monitoring: Systems track temperature, flow rates, and pump performance to maintain reliability and prevent downtime.

Benefits of Liquid-

Cooled Racks

Higher Efficiency: Liquids have higher heat capacity than air, reducing energy consumption and PUE (Power Usage Effectiveness) by up to 40%. Space Savings: Supports more compute per floor tile, enabling 2–2. 5x higher density than air-cooled racks.

Noise Reduction: Eliminates or reduces fan usage, lowering sound levels to near 0 dB in immersion setups. Scalability: Modular designs allow incremental deployment and integration with existing infrastructure. Sustainability: Efficient heat capture enables reuse for facility heating or other applications, reducing water and energy usage.

Considerations for Deployment

Rack Density and Workload: Determine current and future thermal loads, especially for AI and HPC servers with high TDPs. Facility Infrastructure: Evaluate floor load capacity, water supply, and piping routes.

CF

D and flow network modeling help optimize placement. Hybrid Cooling: Combining air and liquid cooling can optimize costs and efficiency, using air for low-density racks and liquid for high-density racks. Maintenance and Reliability: Redundant pumps, hot-swappable CDUs, and leak detection systems ensure continuous operation.

Vendor Solutions: Companies like Supermicro, Flex, Rittal, Schneider Electric, and AIRSYS offer turnkey liquid-cooled racks with integrated CDUs, cold plates, and monitoring systems for AI and HPC environments. Liquid-cooled racks are now a critical component for modern data centers, enabling higher compute density, energy efficiency, and reliable operation for AI, HPC, and other high-performance workloads. Selecting the right cooling architecture—direct-to-chip, immersion, or hybrid—depends on rack density, workload type, and facility capabilities.